Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press RESEARCH COMMUNICATION where groups of genes acquired shared enhancer se- Large scale transgenic and quences acting independently of colinearity. For in- cluster deletion analysis of the stance, the early phase of Hoxd gene expression in limb buds is regulated in a colinear fashion, whereas expres- HoxD complexseparate an sion of the same genes in digits is concurrent, rather than ancestral regulatory module colinear (Nelson et al. 1996). In the HoxD complex, gene recruitment involved in from evolutionary innovations many instances the design of potent enhancer sequences, which regulate several genes at once. We proposed ear- François Spitz,1 Federico Gonzalez,1 lier that expression of four genes in developing digits was Catherine Peichel,2,3 Thomas F. Vogt,2,4 controlled by a unique enhancer that displays poor pro- Denis Duboule,1,5 and József Zákány1 moter specificity as it influenced foreign promoters when targeted to the locus (van der Hoeven et al. 1996; 1Department of Zoology and Animal Biology, University Hérault et al. 1999). Targeted deletions in the posterior of Geneva, Sciences III, 1211 Geneva 4, Switzerland; HoxD complex placed this enhancer somewhere up- 2Department of Molecular Biology, Princeton University, stream of Evx2, outside the cluster (Kondo and Duboule Princeton, New Jersey 08544, USA 1999). Likewise, several genes respond to a gut enhancer sequence that is required to form the ileo-coecal sphinc- The ancestral role of the Hox gene family is specifying ter (Zakany and Duboule 1999) and is localized either in morphogenetic differences along the main body axis. In the first 30 kb of the complex (around Hoxd1) or outside vertebrates, HoxD genes were also co-opted along with of the complex (Kmita et al. 2000b). the emergence of novel structures such as limbs and Regulation inside Hox clusters is complex because of a genitalia. We propose that these functional recruitments high density of genes with embedded and shared regula- tory elements, making it difficult to assign a control se- relied on the appearance, or implementation, of regula- quence to one individual gene, rather than to a series of tory sequences outside of the complex. Whereas trans- genes (Gérard et al. 1996;Hérault et al. 1998;Sharpe et genic human and murine HOXD clusters could function al. 1998). To assess which features of Hoxd gene regula- during axial patterning, in mice they were not expressed tion are intrinsic to the complex and which are located at outside the trunk. Accordingly, deletion of the entire a distance (i.e., act in a global scale over the locus) we cluster abolished axial expression, whereas recently ac- produced transgenic mice carrying additional HoxD loci. quired regulatory controls were preserved. We used a human PAC extending from HOXD3 to 30 kb upstream EVX2 and a mouse BAC containing a tagged Received April 12, 2001;revised version accepted July 2, 2001. HoxD. We compared the regulatory potentials of these transgenic clusters with that of targeted deletions of the During vertebrate development, Hox genes are activated mouse HoxD complex, including a knock-in replace- in a spatio-temporal sequence that leads to partially ment of the cluster. We show that both approaches overlapping transcript domains along the trunk axis. mapped regulatory elements responsible for colinear ex- These expression domains generate various combina- pression within the cluster or close to it, whereas non- tions of HOX proteins at different anterior–posterior po- colinear expression in appendicular structures is dictated sitions, instructing cohorts of cells about their fate (e.g. by regulatory elements located at remote positions. We Krumlauf 1994). In most cases, there exists a correspon- also show that the presence of the complex is required dence between the order of the genes in the genome and for the maintenance of Hox gene expression profiles their domains of expression, a phenomenon referred to as throughout development. colinearity. This feature is very ancient, as it seems to operate in all animals with a bilateral body plan;hence, Results it is likely that colinearity relies on a conserved mecha- Rescue of Hoxd mutations by a human HOXD complex nism. In addition to this function, Hox genes were recruited To study the regulatory potential of a Hox complex, we during evolution to carry out a number of other tasks. produced transgenic mice carrying a 120 kb large human For example, a given subset of vertebrate Hox genes is PAC (Fig. 1A) extending from the HOXD3–HOXD1 re- required for limb development (Davis and Capecchi gion to ∼30 kb 5Ј upstream of EVX2. Five founder ani- 1996;Rijli and Chambon 1997;Zakany et al. 1997), gut mals were recovered. One carried a partial PAC and thus morphogenesis (Zakany and Duboule 1999), or hair fol- was not studied further. Among the others, three lines licle development (Godwin et al. 1998). In some cases, with either one or two copies of the transgene were es- the colinear process was also recruited, whereas else- tablished (TgN[HOXD]1–3). One high-copy number ani- mal was recovered but the line could not be established due to perinatal lethality. [Key Words: Hox cluster;colinearity;BAC;remote enhancers] Present addresses: 3Department of Developmental Biology, Stanford Uni- Mice with the human PAC showed abnormal vertebral versity, Stanford, CA 94305, USA; 4Department of Pharmacology, Merck formulae with a high penetrance of five lumbar vertebrae Research Laboratories, Merck & Co., West Point, PA 19486, USA. (L5) instead of the normal L6 (Fig. 2, top). Besides this 5 Corresponding author. anteriorization, no alteration was observed, suggesting E-MAIL [email protected]; FAX 41-22-702-6795. Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/ that the human genes were expressed during mouse gad.205701. trunk development with the appropriate specificity. This GENES & DEVELOPMENT 15:2209–2214 © 2001 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/01 $5.00; www.genesdev.org 2209 Downloaded from genesdev.cshlp.org on September 30, 2021 - Published by Cold Spring Harbor Laboratory Press Spitz et al. mouse counterparts. Analysis of human transgene ex- pression revealed that the HOXD genes were regulated correctly during trunk development, as expression boundaries were similar to those of the corresponding resident mouse genes (Fig. 3A, top, arrows). Human HOXD13 was weakly expressed posteriorly, starting caudal to the hindlimb bud, whereas expression of HOXD11 was at the level of pv25, around the lumbo– sacral transition. As expected, HOXD4 was expressed up to the hindbrain and in somites, similar to the mouse gene. These results indicated that the rescue observed in the vertebrae derived from a faithful expression of the transgenes. In contrast, expression of human genes in developing limbs was deficient. Early on, weak expression of both HOXD13 and HOXD11 was detected in the posterior limb bud (Fig. 3A, top, arrows). At this stage, transcript distribution differed from that of the corresponding mouse genes, which gave stronger signals over a wider domain including distal parts, in addition to the poste- rior half (Fig. 3A, top). Subsequently (Fig. 3A, bottom), the difference accentuated with an almost complete dis- appearance of all transcripts from the limbs for both HOXD13 and HOXD11, whereas endogenous genes de- veloped their robust expression patterns in both pre- sumptive digits (Hoxd13 and Hoxd11) and forearms (Hoxd11). Expression of human posterior genes was also absent from the developing genitalia, whereas a strong Figure 1. The mouse HoxD cluster, the human PAC, and signal was detected for mouse Hoxd10, Hoxd11, nested deletions. (A)TheHoxD complex (top) with genes as Hoxd12, and Hoxd13. Likewise, expression of the mouse black boxes (red for human). Below is the human PAC 78j1, Hoxd4 and Hoxd11 genes in part of the intestinal hernia with the human HOXD3 to EVX2 genes and 30 kb of DNA (Fig. 3A, bottom, arrows) was not recapitulated by the upstream, and the mouse BAC 400h17, with the murine cluster PAC transgene. and 100 kb of DNA upstream. (B) Scheme of nested deletions. These observations implied that enhancers regulating (Top)TheHoxD complex with a Hoxd11/lacZ reporter trans- Hoxd genes in limbs, genitalia, and intestinal hernia gene inserted upstream of Hoxd13. After recombination of a were absent from the 120-kb human PAC. This was fur- loxP site within Hoxd1 (arrowhead), Cre-mediated deletion re- ther investigated with transgenic mice carrying a 215-kb moved the cluster leaving the reporter gene construct TgHd11/ lacDel9 (below). The bottom lines are intermediate configura- tions obtained by using other loxP sites within the cluster. After recombination, partial deletions of either three (Del3) or seven (Del7) genes are obtained, with the reporter transgene at the same position. was verified further when the transgene was combined with a triple inactivation in cis of Hoxd13, Hoxd12, and Hoxd11 (Zakany and Duboule 1996). Mice (16 out of 18) lacking these functions (HoxDDel3/Del3) displayed an L7 formula (a supernumerary lumbar vertebra;Fig. 2, top). The introduction of the human PAC into this back- ground reinstated the predominant L6 formula (Fig. 2, top) with a substantial incidence of L5 animals. This showed a robust rescue of the mouse defect by the hu- man proteins. Nevertheless, rescue was not detected in the limb skeleton and no difference was scored between wild-type and transgenic hand skeletons (Fig. 2, bottom). Because human HOXD proteins could functionally rescue verte- bral alterations, lack of rescue in limbs was likely due to Figure 2. Vertebral and limb phenotypes. (Top) Skeletal prepa- abnormal expression of the transgenes in these struc- rations of lumbo–sacral transitions. Wild-type mice have six tures. Thus, we analyzed the expression of human lumbar vertebrae (L6).